Figures made from NYT county-level and state-level aggregated data Most recent data: 04/06/2020

  # Note: you need to have the NYT 'covid-19-data' directory in the same parent directory that you're storing the 'covid-analysis' directory that this file is in. otherwise, comment this bit out and input the desired path in the version right below it
county <- read.csv(paste0(gsub(pattern = "/covid-analysis",replacement="", getwd()),"/covid-19-data/us-counties.csv"))
# if specifying path
#county <- read.csv("~/Dropbox/covid-19-data/us-counties.csv")

county$date <- base::as.Date(county$date)
county$state.county <- paste(county$state, county$county, sep=".")

Raw Cases

NYC dwarfs pretty much everywhere in raw number of cases.

log-transformed cases

This squeezes the y-axis (to keep NYC from dominating) and illustrates the exponential growth of the virus (a fixed % growth-rate will be linear in this plot versus exponential in the previous plot)

Also pulling out all counties where a death was one of the first 20 cases reported, to see if they seem to be worse off then counties that either haven’t reported a death or had >20 cases before a reported death. These counties will be in red from here on out

## [1] "We will track these counties in particular:"
## [1] "Mesa, Santa Clara, La Plata, Lane"

It looks like Montezuma County is growing rapidly in a relative sense (growing at ~100% pace every couple days), but hopefully that’s just because numbers are small. -Santa Clara county (Stanford + San Jose) got out of the gates super early, but has been growing at a pretty compared to most other highly infected counties. - La plata county jumped up to almost 30 cases super quickly but is pretty level thereafter. Maybe this is just a testing artifact (backlog cleared in late March, but now building another backlog?) - Mesa County is growing fairly consistently, but still has >40 cases.

If we standardize things by the time when the first case was reported, we can compare the slopes a bit better (except for Sant Clara, which sat at a few cases for almost a month). That other county out by Santa Clara must be King County around Seattle (also doing a pretty good job with slow growth) -Thank god none of us seem to be growing like the real high infection hotspots. Doubling every week is still real scary, but better than doubling every 1-3 days, which is what NYC, New Orleans, and a few other places were doing for a while.

‘THE CURVE’

The epidemiological ‘curve’ of new cases that you’ve no doubt seen everywhere, the thing we’re trying to flatten

This is real noisy, because in most counties we don’t have a ton of new cases per day yet. But one could potentially take heart on the shape of some of these curves. Things might be going down (though the accuracy of recent testing data is highly dubious until we’re a few days out due to backlogs and such).

Same curves, but standardizing the time axis to put all counties on the same timeline:

Same curves but standardized to plot versus days since first case or days since 10th case. Again, maaaaybe a few counties look like they’re peaking? But really hard to say. Let’s look at a different way of assessing this:

Better ‘Curve’ assessment? This is another way of looking to see when exponential growth gets shut down. X axis is total number of cases, y axis is new cases per day (averaged over past 4 days to get rid of some noise, you can still see weekend slumps in test returns though). When new cases stop being a funciton of total cases (i.e. exponential growth stops), the lines will fall off the diagonal. Unfortunately, no one is really doing that yet.

Doing the same thing for deaths rather than cases, we see that, thankfully, we don’t have enough county-level deaths for this to work in most cases.

TRACK YOUR STATE

And last but not least, we can look at state totals just for fun (though the NYT does a better job visualizing this)

Same curves, but standardizing the time axis to put all counties on the same timeline:

New Cases vs Total Cases

No states really seem to be peaking and falling off of the exponential growth phase yet…